This should fix DVB on *BSD.

This supports configurations that use uncommitted diseqc switch
in front of committed diseqc switch to allow 8/16 satellites...

Example of such cascade:

      19.2E  13E    16E  1W      5E 28.2E  23.5E 30W
          \   |      |   /        \   |      |   /
           \1  \2  3/  4/          \1  \2  3/  4/
         +--o---o--o---o--+      +--o---o--o---o--+ LNB (:dvb-satno n)
         | SPAUN SAR 411F |      | SPAUN SAR 411F |
         |    committed   |      |    committed   |
         +-----------o----+      +----o-----------+
                      \              /
                       \1          2/     sat:  switches:
         PORT        +--o----------o--+ 19.2E:  :dvb-uncommitted=1 :dvb-satno=1
(:dvb-uncommitted=n) | SPAUN SUR 211F |   13E:  :dvb-uncommitted=1 :dvb-satno=2
                     |  uncommitted   |   16E:  :dvb-uncommitted=1 :dvb-satno=3
                     +-------o--------+    1W:  :dvb-uncommitted=1 :dvb-satno=4
                             |             5E:  :dvb-uncommitted=2 :dvb-satno=1
                      +------o------+   28.2E:  :dvb-uncommitted=2 :dvb-satno=2
                      | skystar HD2 |   23.5E:  :dvb-uncommitted=2 :dvb-satno=3
                      +-------------+     30W:  :dvb-uncommitted=2 :dvb-satno=4

Reception of 16 satellites is possible using eg. SPAUN SUR 420F as uncommitted
diseqc switch instead of SPAUN SUR 211F, and 2 more SPAUN SAR 411F committed
diseqc switches.
Signed-off-by: Rémi Denis-Courmont <remi@remlab.net>

Also list the entries in the log like ALSA output.

FE_DISEQC_SEND_MASTER_CMD will fail if command length is not set,
preventing multi-satellite tunning using DiSEqC switch.
Signed-off-by: Rémi Denis-Courmont <remi@remlab.net>

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

In some cases, that gets more accurate timing meaurements faster.

VLC does not use pa_stream_get_time(), so there is certainly no need
for pa_stream_get_time() to return monotonic values.

Such a latency update is an opportunity to reschedule the stream
trigger time, or to trigger immediately if late. That can reduce the
initial audio delay in some cases.

Also remove unchecked memory allocation.

-O3 enables dangerous optimizations and does not make much sense for a
protocol (not decoding => not performance critical) library.

Close #7133

Close #6578

The code is far from being great, but it matches the Qt intf now.

Fix #1859 #6983

Depending on the system, this can be called ducking, focus, policy...
But it basically boils down to pausing playback automatically when a
higher priority audio stream starts, typically a voice call.

Resuming playback automatically when communications are over would be
possible as well. But that seems unfriendly to me, so I skipped it.